Survey of Test Methods for Convolutional Neural Network Object Detection in New Environments

Author(s):

Craig Lennon

Abstract:

This report surveys and provides demonstrations of techniques that could be applied to test a convolutional neural network for application in a new environment using the Ultralytics YOLOv8n model. Army programs using such a network for object detection in an environment for which they lack sufficient training and test data can consider this survey to see if any of the methods demonstrated here have potential value to their program.